![]()
Abstract The estimation of reservoir properties is a risky issue. Uncertainties come around; the number of wells (hard data) is often low and poorly distributed. Densely sampled, 3D seismic information (soft information) has been seen as a key to reducing interwell uncertainties. As it is well known, however, even after seismic processing, noises and sign distortions remain if not introduced by the processing itself. Moreover, inconsistency in horizons picking, during interpretation, worsens the quality of the obtained attributes with corresponding impact for its use as a predictive variable for reservoir mean properties estimation.
In this paper, we discuss incorporation of seismic attributes filtered by Factorial Kriging via conditional simulation on reservoir porous volume estimation. The seismic attribute mean acoustic impedance obtained from a 3D seismic program with seismic stratigraphic inversion is considered. It shows itself to be well correlated to the porous thickness of the reservoir interval under consideration. A nested corregionalized model, which presented two structures, was fitted to the attribute. The small range associated to data noises was filtered by Factorial Kriging, which improved the correlation between the acoustic impedance and the porous thickness. The dispersion of the conditional simulation constrained by the filtered seismic attribute is lower than the dispersion of the conditional simulation constrained by the pre-filtering attribute, reducing the uncertainty of the reservoir porous volume estimation.
Introduction The integration of well data and seismic attributes using geostatistics has, nowadays, become more popular. Due to the limitations imposed by the seismic resolution, the reservoir mean proprieties are not correlated to seismic attribute volume, but to seismic attribute mean maps. It is well known, however, that, even after seismic processing, noises and sign distortions remain if not introduced by the processing itself. Moreover, inconsistency in horizons picking, during interpretation, worsens the quality of the obtained attributes with corresponding impact for its use as a predictive variable for reservoir mean properties estimation. It is also worth mentioning that monoattribute or mean maps may disguise the image of geologic features of different dimensions, hard to distinguish through conventional filtering techniques.
In this paper, we discuss incorporation of seismic attributes filtered by Factorial Kriging via conditional simulation on reservoir porous volume estimation. The next section provides the theoretical background assumed. Section 3 brings considerations about the data set used for the analysis. The attribute images before and after Factorial Kriging filtering are displayed in Section 4 as well as the description of the impact of its use for the reservoir volume estimation. Section 5 is the conclusion.
Technical Background Factorial kriging works in the spatial domain in a similar way to the spectral analysis in the frequency domain. Events which can not be distinguished in the Fourier transformed domain may be separated in a variographic analysis and filtered by factorial kriging.
Seismic attributes, contrary to well data, are densely sampled, and its integration in kriging or conditional simulations systems enhances the interwell estimations. The decision about the seismic attribute feasibility is, in general, based in the correlation analysis between the attribute and the well data. Good correlations allow the attribute use. Nevertheless, applications to improve the attribute quality are not pursued.
Site News